Hang in mousedev_init

Hang in mousedev_init

[Eric Paris]

I have a report of a person who is booting a 2.6.25 based fedora kernel
and hanging in mounsedev_init().  Moving to our old 2.6.24 based fedora
kernels he doesn't see any problems.

we know its mousedev_init since we added initcall_debug and
mousedev_init() is called but never returns...

Doesn't look like many patches between then and now

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=3b04a61107dfe46dbfc1796298b59ca3c0a09cd9
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=d182c10c842007984e12b3b816df2b10d997cc8e
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=a7097ff89c3204737a07eecbc83f9ae6002cc534

About the only thing my clueless eye noticed was in
d182c10c842007984e12b3b816df2b10d997cc8e

--- a/drivers/input/mousedev.c
+++ b/drivers/input/mousedev.c
@@ -1032,7 +1032,7 @@ static const struct input_device_id mousedev_ids[] = {
                .flags = INPUT_DEVICE_ID_MATCH_EVBIT |
                        INPUT_DEVICE_ID_MATCH_KEYBIT |
                        INPUT_DEVICE_ID_MATCH_ABSBIT,
-               .evbit = { BIT(EV_KEY) | BIT(EV_ABS) | BIT(EV_SYN) },
+               .evbit = { BIT_MASK(EV_KEY) | BIT_MASK(EV_ABS) },
                .keybit = { [BIT_WORD(BTN_LEFT)] = BIT_MASK(BTN_LEFT) },
                .absbit = { BIT_MASK(ABS_X) | BIT_MASK(ABS_Y) },
        },      /* Mouse-like device with absolute X and Y but ordinary

You dropped the | BIT(EV_SYN)   does that matter?  What could have
changed to cause a hang?  Any clues?

Re: Hang in mousedev_init

[Dmitry Torokhov]

On Wed, May 28, 2008 at 11:49:42AM -0400, Eric Paris wrote:
> I have a report of a person who is booting a 2.6.25 based fedora kernel
> and hanging in mounsedev_init().  Moving to our old 2.6.24 based fedora
> kernels he doesn't see any problems.
> 
> we know its mousedev_init since we added initcall_debug and
> mousedev_init() is called but never returns...
> 

Is it possible to get a SysRQ-P and SysRq-T traces and see where we
get stuck? Does the boot progress if you exclude mousedev from the
kernel?

> Doesn't look like many patches between then and now
> 
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=3b04a61107dfe46dbfc1796298b59ca3c0a09cd9
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=d182c10c842007984e12b3b816df2b10d997cc8e
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=a7097ff89c3204737a07eecbc83f9ae6002cc534
> 
> About the only thing my clueless eye noticed was in
> d182c10c842007984e12b3b816df2b10d997cc8e
> 
> --- a/drivers/input/mousedev.c
> +++ b/drivers/input/mousedev.c
> @@ -1032,7 +1032,7 @@ static const struct input_device_id mousedev_ids[] = {
>                 .flags = INPUT_DEVICE_ID_MATCH_EVBIT |
>                         INPUT_DEVICE_ID_MATCH_KEYBIT |
>                         INPUT_DEVICE_ID_MATCH_ABSBIT,
> -               .evbit = { BIT(EV_KEY) | BIT(EV_ABS) | BIT(EV_SYN) },
> +               .evbit = { BIT_MASK(EV_KEY) | BIT_MASK(EV_ABS) },
>                 .keybit = { [BIT_WORD(BTN_LEFT)] = BIT_MASK(BTN_LEFT) },
>                 .absbit = { BIT_MASK(ABS_X) | BIT_MASK(ABS_Y) },
>         },      /* Mouse-like device with absolute X and Y but ordinary
> 
> You dropped the | BIT(EV_SYN)   does that matter?  What could have
> changed to cause a hang?  Any clues?
> 

Every device has EV_SYN so it makes no sense to try to match on it.

-- 
Dmitry